Just by accident found this piece on transcribing podcasts with IBM's Watson. I wonder how it compares with SwiftScribe? Need to give it a try, because as I attempt more complex stories it certainly helps to have a bit of text in addition to notes and markers on the audio.